Michael Darius Eastwood · Independent AI alignment researcher
Published
The Polymathic Neurodivergent Profile, or PNP, is a descriptive framework, not a diagnostic category. Working paper C of the ARC/Eden research programme proposes it for a pattern that existing psychiatric nosology, disability law, and educational classification do not cleanly capture. The paper lists six components and grades each by evidence strength. Five are rated STRONG. One is rated WEAK, and the paper says so on the first page. That honest grading is the reason for publishing the framework in the first place.
The components are: cross-domain learning, monotropic hyperfocus, executive-function asymmetry, self-directed research intensity, institutional misrecognition, and the descriptive label "polymathy" itself. The first is the acquisition of competence across multiple fields at unusual speed, with structural pattern transfer between them. The second is sustained, intense, self-directed attention in high-interest domains, producing output disproportionate to formal training. The third is the marked discrepancy between domain-specific high performance and routine administrative, temporal, and organisational functioning. The fourth is the capacity to initiate, sustain, and complete large-scale independent projects without institutional scaffolding. The fifth is the systematic failure of institutions to identify, accommodate, or appropriately evaluate the profile. The sixth is the "polymath" label used descriptively.
Monotropic hyperfocus rests on Murray, Lesser and Lawson's 2005 attention-tunnelling account of autism, and on hyperfocus work by Russell and colleagues (2019). Executive-function asymmetry rests on NICE CG142's explicit direction that clinicians document the IQ and adaptive-functioning gap in autistic adults, and on the twice-exceptionality literature (Foley-Nicpon, Assouline and Colangelo, 2013; Reis, Baum and Burke, 2014). Institutional misrecognition rests on the autistic underemployment, masking, and burnout literatures (Cage and Troxell-Whitman, 2019; Raymaker and colleagues, 2020; Higgins and colleagues, 2021). Cross-domain learning and self-directed research intensity are observational rather than experimentally quantified, but they are consistent with the polymathy literature (Root-Bernstein, 2009; Araki, 2018) and with special-interest depth work in autism (Grove and colleagues, 2018).
"Polymath" is a descriptive label with no standing in psychiatric nosology, disability law, or educational classification. The working paper retains it because the existing diagnostic vocabulary captures fragments of the pattern (giftedness, special interest, high IQ) but not the specific combination of extraordinary cross-domain range with clinically significant executive-function asymmetry. The paper does not upgrade the label. It prints the WEAK grade in the same table as the strong ones. Anyone reading the paper is told exactly how strong each component is before deciding whether to apply the framework.
PNP is falsifiable by construction, and the paper says so. The framework must add incremental predictive value beyond existing autism, ADHD, and IQ formulations. If those already fully account for the variance PNP claims to explain, PNP collapses into them and must be retired. It must be operationally distinguishable in principle from ordinary high openness, ordinary giftedness, narcissistic self-description, and generic executive dysfunction. And its concrete institutional prediction, that visible high performance produces systematic under-recognition of support needs, must be demonstrably distinct from the pattern predicted by any single existing framework alone.
From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.